Shot Selection Round-Up: Part 2
Over the past week and a half, I’ve been slowly carving out time to go back and review my season-long shot selection numbers, using the metric Expected Points Per Shot (XPPS). If this is your first time reading about XPPS, here are the basics:
Shots from different locations are provide different value. For example, a layup has higher chance of being made than a long two-pointer, and a three-pointer earns an extra point. We refer to these different values as the expected value of a shot. XPPS looks at all the shots a player or team takes and boils that down to one average expected value per shot. Free throw attempts are included as well, so from here on out when I refer to shot attempts, I’m referring to true shot attempts (field goals and trips to the free throw line). In the end, more shots at the rim, free throws and three-pointers means a higher XPPS. More long two-pointers sends the number in the other direction.
I’ve been keeping track of the numbers all season long for individual players, teams and team defenses, all with interactive visualizations. Last week we looked at some of the best and worst shot selections this season from individual players. Today I want to dig into a few of the most and least impressive team numbers.
This season’s worst shot selection belongs to the Philadelphia 76ers, with an XPPS of 1.016. League average this season was 1.047. The culprit was the ugliest weapon in any offensive arsenal, the mid-range jumpshot, making up 32.0% of their true shot attempts. As a team they attempted nearly a thousand more long two-pointers than three-pointers. That sort of reliance on the mid-range shot certainly hinders offensive efficiency, but it doesn’t outright prohibit it. However, in addition to taking a boatload of long two-pointers the 76ers weren’t especially good at making them. They outperformed their XPPS by just 0.003, for an actual points per shot of 1.019, the second worst mark in the league.
Again, shot selection is not the be-all, end-all variable for good offense but its impact is significant. In a hypothetical world where the 76ers’ shooting percentages from each area of the floor where completely static and immune to changes from outside factors; bringing just their shot selection up to league average XPPS would raise their TS% from 50.9%, 29th in the league this season, to 51.7%, which would have ranked 21st. Using that number, some regression analysis I’ve done previously, and holding their OReb% and TO% constant, we can predict that their ORtg. would jump from the 99.9 they actually averaged in the regular season to 102.9. That would essentially give them an even point differential for the season, the mark of a 0.500 team. I know I’m sprinting down a slippery slope here, assuming a huge number of other variables wouldn’t change as well, but since the 76ers finished at 34-48 we can estimate that their shot selection may have cost them as many as seven wins this season.
With new GM Sam Hinkie on board there are sure to be some foundational changes to how the 76ers go about their business. I’m certainly not the first one to point it out, but working on shot selection may be a good place to start.
At the other end of the spectrum we find a pair of teams working shot selection to some unbelievable advantages. The top two team shot selections this season, measured by XPPS, were the Houston Rockets (1.103) and the Denver Nuggets (1.094). Those two marks were not only the highest this season, they’re the two highest for the last 12 NBA seasons (which is as far back as my data goes). Given the way shot selection is trending in the NBA, it may be safe to say that they’re the two most efficient shot selections of the three-point era.
Although they were both on the extreme ends of efficiency in terms of offensive decision making, the Nuggets and Rockets went about things in different ways. Both teams abhorred mid-range jumpshots, and their percentage of shot attempts used on long two-pointers, 12.9% for the Rockets and 15.8% for the Nuggets, were the lowest for the past 12 seasons. But in avoiding those inefficient shots, they went in two directions. 30.8% of the Rockets true shot attempts were three-pointers this season, the second most in the league behind the Knicks. Meanwhile, 41.9% of the Nuggets’ shot attempts came at the rim, by far the best percentage in the league. The second best mark in the league did belong to the Rockets, at 35.5%. So while the Rockets avoided inefficiency primarily by retreating behind the three-point line, the Nuggets headed for the rim at unprecedented rates.
Each team used ultra-efficient shot selection to compensate for a different weakness. The Nuggets a lacked a single player with enough offensive firepower to unilaterally destroy a defense. What they had was a collection of talented, if slightly more limited offensive players, who’s abilities they maximized by hammering them into that ultra-efficient framework. The Rockets had that uniquely talented individual scorer in James Harden, but lacked the offensive depth that the Nuggets have. For them efficient shot selection was a way of scaffolding their limitations around Harden. In both cases the results were extremely successful as the finished the regular season as the 6th and 7th most efficient offenses in the league.
I did one other piece of analysis with these two team’s shot selections numbers, sparked by Daryl Morey’s comments about trying to increase variance in their playoff series against the Thunder by taking more three-pointers. Although three-pointers are an incredibly valuable shot because of the extra point each make earns, they’re made less frequently. In a large sample the inaccuracy and added value even out over time, making them worth the risk. But in a small sample size, three-pointers introduce a lot of instability to offensive output because of the reduced rate at which they put points on the board.
Although the quality of their shot selections and offensive efficiency were very similar, the fact that the Nuggets relied so heavily on interior shots made their offense more stable than the Rockets. I looked at the game-by-game offensive efficiency for each team this season and measured the variance. The Rockets’ variance in ORtg. worked out to 177.3, nearly twice as large as the Nuggets’ 95.7. That means that although the Rockets’ offense produced higher highs this season, it also produced lower lows. The reliance on the three-pointer made their offense less consistent on a night-to-night basis. If you’ve ever heard an announcer say a team “lives by the three, and dies by the three” they’re referring to this phenomenon in slightly more theatrical terms. Exploring this link between shot selection and variance in offensive efficiency is definitely on my summer to-do list.
Check back in next week and we’ll review the shot-selection numbers for team defense. Spoiler alert – there are some historic numbers to be found there as well.
In case you missed the link further up the page, all the team numbers I referenced can be found here.